The Pittsburgh Post-Gazette on how the world promised by AI isn’t necessarily a better one:

Artificial intelligence is reshaping human beings’ relationship with the world around us AI offers, for the first time, the possibility of human obsolescence in the capacity that most defines us: our intelligence.

The benefits and dangers of AI could each be much greater than for any previous technological development. Indeed, we believe the dangers are likely to outweigh the benefits. For this reason, we urge caution in the further development of AI, strict application of safety standards and generous use of government regulatory power.

Further, while the proliferation of AI is at this point all but impossible to contain, we urge individuals and institutions, as far as possible, to minimize its use and to preserve pre-AI ways of learning, of communicating and of executing tasks. The human mind has the quality of a muscle, and it will atrophy if its higher-order responsibilities are outsourced to machines.

At the same time, understanding that the further development of AI is inevitable, but this includes AI leaders making binding commitments to the highest standards of safety and ethics.

The San Francisco-based Center for AI Safety has developed a single-sentence “Statement on AI Risk,” which reads: “Mitigating the risk of extinction from AI should be a global priority alongside other societal-scale risks such as pandemics and nuclear war.” Hundreds of the most prominent and prolific AI researchers, tech moguls and policymakers have signed this statement. We agree that the civilizational risks are real.

But we’d also like to focus on more immediate concerns. Large-language models, for instance, are trained on billions of words written by human beings who have not been, and likely never will be, compensated for the use of their labor. The deployment of AI in educational settings — much more than previous developments in information technology — quickly leads to reliance on often-faulty models as opposed to the intellectual exercise of searching for, consuming and synthesizing ideas and information.

Further, the amount of energy needed to power AI datacenters and supercomputers is basically ending any hope of achieving a sustainable equilibrium between human civilization and the natural world. The dream of an all-renewable energy portfolio is dead. While we favor focusing on the expansion of nuclear power generation, the current AI energy-consumption trajectory will make the expanded use of fossil fuels impossible to avoid.

These are but of a few of the serious threats to overall human thriving posed by AI. Whether the promised benefits can outweigh them remains speculative, at best.